A computer-assisted medical decision-making (CMD) system is an interactive computer system which can directly assist physicians with a clinical decision-making task. Several of these systems have been developed for use in neurologic, problems, with variable success. These systems have generally not been good models of human diagnostic reasoning, have had difficulty with the coexistence of multiple diseases and have not been capable of sophisticated anatomic reasoning. Adequate assessment of functioning programs has proven difficult to accomplish and has rarely been performed. Physician acceptance of these systems has been limited by these and other factors which this research proposal seeks to address. In the first phase of the proposed research, an existing rule-based CMD system for TIA diagnosis and management will be incorporated into a comprehensive system, designed to aid the physician in the diagnosis and management of patients at risk for impending stroke in general. This system will be tested using records from 100 patients undergoing non-invasive carotid artery evaluation. Its performance will be compared to that of a stroke specialist and the physicians caring for the patients. The second phase of the proposed research will focus on the development and testing of a computer program for neurologic localization. Unlike previous programs this will have a well grounded theoretical basis in set cover theory, will employ sophisticated spatial inference and cause-effect reasoning and be capable of answer justification. Once developed and tested, this program will be incorporated into CMD systems for neurologic diagnosis and management. The completion of a system with these capabilities will represent an advance in the field of artificial intelligence as well as be of benefit to practicing clinicians and in the education of future physicians.